Log of SALT-RSS observations

Spectral classifications of 49 objects are available in the literature, while 32 objects are newly classified. The known spectral types were used to test an automated classification scheme, which relies primarily on template fitting, supplemented by matching of spectral indices calibrated against the template spectra.

An attempt was made to quantify the uncertainty in the spectral types, which is generally better than two subclasses. Objects for which spectral types differ by more than one subclass from the literature classifications are discussed individually. Discrepancies between automated classifications based on, respectively, template fitting and spectral index matching, may be useful for flagging objects with unusual spectra.